Response to jack Waugh on 'hard'cmavo (old posting)

[lojbab (Bob LeChevalier)]

On Tue, 5 Mar 91 10:47:08, uunet!mcnc.org!aurs01!aurw31!waugh (Jack Waugh)
wrote:

>What I meant by "hard" little words was ones that the listner would have
>to know in order to parse a sentence.  Let me refine that proto-proposal
>a little (forget about "hard"):

>An adequate model to use for the first stages of listening would be to
>say that the listner parses first and then starts semantic processing.
>Parsing is casting the utterance into a tree, so as to know the grouping
>relationships.  Once this is done, the listener turns to the semantics
>of the words.  ...

>Of course, the parsing rules for Lisp would not be appropriate for a
>spoken language for use between humans.  However, maybe we could design
>a set of conventions which would require the parsing operation to act on
>less information than the whole cmavo vocabulary, just as a Lisp parser
>doesn't have to know the behavior of the functions.

There is a small error in Jack's presentation of the 'adequate model' for
the first stage of learning, and two obvious erroneous assumption as well.

The obvious erroneous assumptions are that the speaker is uncooperative
and will use words that she/he knows the listener won't likely understand.
And second, that the speaker gives no clues to syntax in phrasing and
tone (especially when talking to novice learners).

Both are untrue by our experiences here in Lojban conversing each week,
now that we've had a few weeks to see what happens.  We have people at
every level of ability from true novice to me, who seldom needs a word
list.  There is a lot of repetition (not to mention self-correction.
When speaking to someone less skilled, there is a tendency to talk
slower, to speak sumti as separate phrases, and to convey other parse
features in speech.  I've since notivced that we do the same in English,
even speaking fluently to other English speakers, especially when there
is a 'noisy environment' that might cause individual words to be
missed/misunderstood.

The latter is the clue to how Lojban learning speech is parsable, and
hence the error in the model presentation.  Regardless of how the
speaker speaks, REAL human listeners do not parse one word at a time.  I
suspect that we recognize an entire phrase as a unit, and to the
extent I can conciously analyze my own practice, I certainly process
Lojban phrases as a unit.

Lojban actually has only a couple of phrase structures to learn - sumti,
selbri, bridi, and maybe eventually MEX, but that isn't a problem for a
new learner.  There are also attitudinals, and they can gum up the
parsing if used irregularly, one thing that makes Michael Helsem's usage
almost unintelligible.  But he isn't a cooperative speaker.  Normally,
attitudinals and discursives are used at the beginning of the sentence,
or there is emphatic stress because they are occurring in an unusual
place.  Three important structures to learn to recognize, each with only
a few internal patterns.  You learn to recognize the patterns as a
whole, and a cooperative speaker will help by using phrasing breaks
bewteen each structure.  The role of the words then is implicit in
how they interact in the structure.

   A brief example, of selbri only.  The following are all the basic
   patterns I think a learner needs to recognize for beginning AND
   intermediate usage.  Perhaps a dozen cmavo need to be learned, learn,
   plus the logical connectives which are a series.  Anything more is not
   'appropriate' for a learner.  The key mark of the selbri is the initial
   brivla (or possibly one of the abstractors). brivla have consonant
   clusters - they stand out morphologically.  The other key words to learn
   are se/te/ve/xe converters (which are a series in alpha order) nu/ka/ni
   abstractors and terminator kei, bo/ke/ke'e groupers/terminator and co
   inverter. me (and seldom needed terminator me'u) converts a sumti to a
   brivla-like grammar; but the me means that it is a brivla - so you know
   how to parse it.  After a brivla, be/bei/terminator be'o glue a sumti
   into a selbri without terminating it.

   Key patterns of selbri components, each with approximate grammar of a brivla:
        brivla
        brivla + brivla
        brivla + co + brivla
        brivla + bo + brivla
        brivla + je-series-connective + brivla
        ke + brivla + [ke'e]
        nu/ka/ni + brivla/bridi + [kei] (I'll come back to this)
        me + sumti + [me'u]
        brivla + be + sumti + [bei + sumti]... + [be'o]

Most if not all of these patterns are described in the 7-page diagrammed
sentence examples that are sent to everyone, even level 0 people in the
introductory packet.  I believe old-timers got a copy around JL/LK
issues #10 or so.  These help you speak and to understand Lojban.  Nick
Nicholas recently sent me a nearly perfect Lojban sentence on his first
attempt, after about 3 hours reading the level 1 materials and JL13.
This wasn't a simple sentence either, using a non-standard sentence
order and a nu abstraction.  While there is no index for vocabulary list
for this tiny document, there can't be more than a few dozen cmavo in
it.  These are the most important ones to learn.  Then (to specifically
answer Jack's comment), you then learn to recognize the grammatical
function of the others by seeing how they modify these basic structures.

In listening, the experience of the Lojban conversationalists is that
recognizing grammatical structures is NOT a significant problem compared
to learning the gismu list so you don't have to look every word up.  If
you did what we said and used LogFlash/flashcards/some other technique
to learn the gismu vocabulary 3 years ago or whenever you got
interested, you would be able to understand Lojban conversation within
perhaps an hour or two, with little difficulty.  The cmavo can be
studied systematically, but the important ones are picked up by
seeing/hearing them used.

Speaking/writing is a bit more difficult because you have to know the
structures well enough to make choices amnong them; I'll address it at
another point, since it wasn't Jack's question.  But briefly, the real
problem here is that novices think in English and then want to construct
Lojban sentences from the start with the full power and complexity of
English sentences.  You have to walk before you can run.  Decide the
key relationship you are trying to express, pick a gismu or tanru
that expresses it, and then attach sumti representing the ideas
you are trying to relate.  If you can't decide one single relation
that you can express in a gismu or short tanru, you are trying to say
something too complex for a beginner.

Reading is also more difficult.  A writer like myself who knows the
vocabulary better is going to use more of them, and more complex
structrues.  We also have the time to check over what we say to be sure
it makes sense (Unlike the stuff posted on this list by me and others,
stuff that goes into JL or LK is much more thoroughly checked for errors
and clarity.  You won't likely learn the language just reading this list,
and certainly you won't get the best examples.)

I'll let John Cowan or anyone else pose some specific examples, perhaps
from his diary or some other simple text, showing how beginner's parse
or build sentences in practice.  Or you can look at the diagram sheet.
But as I said above, it really is easy.

On one other point:  Jack seems to want words in series, and with
morphological similarity.  Of course we have the basic division between
gismu cmavo and names built into the morphology.  Within cmavo, also,
the V and VV words are morphologically distinct, and several of the more
important clusters of words are series, but that tenses modals and
discursives are not.  These latter are particularly numerous, and we
decided it was more important to make them easily recognizable in terms
of a gismu with related meaning than to make an artificial series that
wouldn't be memorable.

Putting word in highly memorable series sounds like a great idea in
aiding memory and parsing, but in reality, it is not a benefit.  We even
moved the numbers OUT of a simple series into a more complex one because
the decrease in redundancy from having just one common letter makes the
language HARDER to understand in real usage.  We are having more
problems with people mixing the grammatically identical le and lei (due
to not knowing how to pronounce Romance 'e') than with any other
pronunication problem, including the English tendency to turn unstressed
vowels into schwa (y), which is one of Jack old-time complaints.

As with many other things in the language design, learning at first
seems harder than it should be because you have to work at memorizing
more words.  But once you have learned them, the language seems quite
simple.

-lojbab